Autonomous Earth-Moving Vehicle Learning for Changing Jobsite Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing earth-moving vehicles (EMVs) are dangerous, costly, and inefficient due to reliance on human operators and static machine learning models that do not adapt efficiently to changing environments, leading to accidents and prolonged training times.
Innovation Solution
Implementing an augmented learning system with two machine learning models - a world model and a behavior model - that are fine-tuned using real-time sensor data to enhance the efficiency and adaptability of EMVs, allowing for autonomous operation and reduced training times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human operators control EMVs, then operational flexibility and decision-making are maintained, but safety risks and operational costs increase
Solution Approach 1:
The EMV system performs self-learning and self-improvement through continuous fine-tuning of machine learning models using real-time sensor data and outcome feedback, enabling autonomous operation without human intervention while improving safety and reliability
Solution Approach 2:
The system implements closed-loop feedback by processing outcomes of actions performed by the EMV and using this information to fine-tune the machine learning models, enabling continuous improvement of autonomous operation safety and effectiveness
2Adaptability or versatility
If static machine learning models are used for EMV operation, then system complexity is reduced, but adaptability to changing environments deteriorates
Solution Approach 1:
The machine learning models transition from static to dynamic through continuous fine-tuning processes that adapt model parameters based on real-time sensor data and environmental conditions, enabling the system to respond to changing environments while maintaining manageable complexity through automated learning
Solution Approach 2:
The system performs preliminary training of machine learning models on historical sensor data before deployment, and continues fine-tuning with real-time data, preparing the models in advance for various environmental conditions and reducing the complexity of real-time decision-making
3Productivity
If traditional training methods are used for machine learning models, then training completeness is achieved, but training time increases
Solution Approach 1:
The fine-tuning process operates continuously in the background during EMV operation, utilizing idle processing cycles to update models with real-time sensor data, thereby maintaining training efficiency without interrupting operational productivity
Solution Approach 2:
The system applies partial fine-tuning by selectively updating only the most critical model parameters based on current operational needs and outcome feedback, rather than retraining entire models, thus reducing training time while maintaining sufficient model performance
Data Source
AI summary
Systems and methods for using augmented learning models for autonomous earth-moving vehicles are disclosed. The method can comprise receiving a second set of sensor data; generating a first condensed vector from the second set of sensor data at least in part by processing the second set of sensor data with a first machine learning model; selecting an action to be performed by the vehicle at least in part by processing the first condensed vector with a second machine learning model. The method can further comprise retrieving one or more samples of sensor data from the first set of sensor data; fine-tuning the first machine learning model at least in part by processing the one or more samples of sensor data to produce a second condensed vector; and fine-tuning the second machine learning model at least in part by processing the second condensed vector.